What’s the Worst Length for an Instagram Reel? A data led study

On Instagram, do you make your videos short to protect reach – or let them run long enough to say something useful? It is a decision every social media team makes, usually on instinct rather than evidence. So we decided to take a look at the data. 

Over almost three years we posted 68 Instagram videos ranging from 21 seconds to just over five and a half minutes, covering everything from client results to office celebrations. Here is what the numbers say about how length actually performs, and where it stops mattering as much as the story being told.

Reach falls steadily as videos get longer

Across the full sample, video length and view count move in opposite directions. The correlation between duration and views is -0.38: a moderate but consistent relationship in which longer Reels are watched by fewer people.

The one to one-and-a-half minute bracket is the clear reach outlier, averaging 463 views against an account average of 263. Every bracket after that trends downward until length passes three minutes, where views bottom out at 125 to 142.

Table 1: Average performance by video length

Length	Videos	Average views	Average engagement rate
Under 1 minute	11	303	4.1%
1 to 1.5 minutes	7	463	3.6%
1.5 to 2 minutes	9	286	3.2%
2 to 2.5 minutes	14	276	2.8%
2.5 to 3 minutes	12	237	2.9%
3 to 3.5 minutes	9	142	6.5%
3.5 to 4 minutes	3	125	5.2%
Over 4 minutes	3	133	6.6%

Figure 1: Average views by video length (n=68)

The 2-3 minute zone is the weakest length on Instagram

Engagement rate does not fall in a straight line with length. It follows a U-shape: strongest at the short end, weakest in the middle, and strongest again once a video runs past three minutes. Videos under one minute average a 4.1% engagement rate. Videos between two and three minutes average 2.8% to 2.9%, the lowest in the dataset. Videos over three minutes average between 5.2% and 6.6%, roughly double the middle bracket.

This is the most interesting, and the most counter-intuitive, finding in the data. A two-to-three-minute video looks like the safe choice. It’s not so short it feels lightweight, and not so long it risks drop-off. But on Instagram, it is neither. It loses on reach and gives up more engagement than either extreme. Fourteen of the 68 videos in this sample, the largest single length bracket, sit in that middle zone.

Figure 2: Average engagement rate by video length, against the account average of 3.9%

Long-form videos filter for intent, not for reach

Past three minutes, the audience narrows sharply but engages far harder with what remains. A video explaining why AI-generated content is not ruining content marketing ran 3 minutes 16 seconds and reached only 80 people, but one in ten of them interacted with it. A five-tip ecommerce checklist video ran 3 minutes 2 seconds, reached 110 people, and saw the same one-in-ten interaction rate. Both sit among the five highest engagement rates in the entire dataset, on some of the lowest view counts.

This pattern suggests long-form videos stop working as a discovery format and start working as a filter. Casual scrollers drop off before the two-minute mark. The people who stay are the ones who came specifically for the advice, and they behave like it. A brand purely chasing reach should keep videos short. A brand trying to prove specific expertise to a smaller, more qualified audience has a legitimate case for running long.

Content type shapes audience size and appetite

Length is not the only lever. What a video is about changes who shows up, before length changes how hard they engage once they are watching.

• Culture and behind-the-scenes content – such as office celebrations, taste tests and team events – nearly doubles the account’s average reach. 
• Practical tips content has the opposite profile: the lowest average views of any category, but the second-highest engagement rate. 
• Predictions and trend commentary earns the single highest engagement rate in the dataset, on modest reach. 
• Client results and case studies, the account’s most-produced format at nearly a third of all output, sit almost exactly on the account average for both measures – a reliable workhorse.

Table 2: Average performance by content type

Content type	Videos	Average length	Average views	Average engagement rate
Culture, events and behind-the-scenes	7	63 sec	498	4.0%
Client results and case studies	22	156 sec	229	3.5%
Predictions and industry trend commentary	6	136 sec	217	5.4%
Team profiles and introductions	5	191 sec	210	4.4%
Practical tips and how-to advice	7	174 sec	166	4.7%
Other (offers, news and general updates)	21	104 sec	280	3.6%

Figure 3: Average views (bars) and engagement rate (line) by content type

One caveat belongs here. In this dataset, content type and length are not independent variables. Culture content is almost always short. Practical tips and predictions content is almost always long. So it is not yet possible to say definitively whether topic or length is doing the work when a behind-the-scenes video and a practical-tips video perform differently, because the two currently move together throughout the account’s output. 

What can be said with confidence is that content type shapes audience size and appetite. It changes who is willing to watch, even if separating the exact contribution of topic from duration would need a genuine test, such as a longer culture video or a sub-minute tips video, run against this baseline.

What this means for social media managers

For teams making these decisions on a content calendar rather than in a spreadsheet, four things follow from this data.

1. Decide reach or depth before you brief the edit

Before cutting a video, decide what job it needs to do. A video meant to introduce the brand to as many people as possible should sit under 90 seconds. A video meant to convert a smaller audience already looking for the answer, such as detailed advice or a client result, can justify running past three minutes. Treating every video as though it needs to do both jobs at once is how content ends up in the weakest length bracket by accident.

2. Retire the 2-3 minute default

This length is the single most common runtime in this dataset and the weakest performer on both counts. If a script naturally lands here, it is worth asking whether it can be tightened toward 90 seconds, or whether the topic genuinely needs the extra depth that pushes it past three minutes. There is little upside to stopping in the middle.

3. Pair content type and length deliberately, then test the pairing

Short-form Reels are the natural home for behind-the-scenes and culture content, where the priority is reach. Longer-form Reels are the natural home for advice and predictions content, where the priority is a smaller, more committed audience. Testing the reverse, a longer culture video or a sub-minute advice clip, would help separate how much of each result is down to topic and how much is down to length.

4. Redefine what a good video looks like by objective

A video with 130 views and a 6% engagement rate is not underperforming. It is doing a different job to a video with 460 views and a 3.5% engagement rate. Reporting both against the same reach target misreads the data and risks cutting the exact content that is earning the most attention from the audience it was made for.

Taking a balanced approach

None of this points to a single optimal length. It points to a deliberate mix. Short videos under 90 seconds should carry the content built to travel: culture, announcements, and moments worth a broad audience seeing. Videos past three minutes should be reserved for content built to convert a smaller audience already paying attention: detailed advice, predictions, and client results that need room to be specific.

We asked a similar question of our LinkedIn channel, and the trade-off between reach and depth showed up there too, in a different shape for a different platform and audience. The two-to-three-minute length should not disappear from an Instagram content calendar by default, but it should only appear when a video genuinely cannot be shorter and does not yet need to be longer, rather than as the resting length every script drifts toward.

Final thoughts

The instinct to keep every video short protects reach at the cost of depth. The instinct to let every video run long protects depth at the cost of reach. Neither instinct holds up against our own posting data. Length is a decision with a real trade-off attached, and this account’s own performance shows exactly where that trade-off lands.

Methodology

Data scope and duration. This analysis covers all 68 video posts published on The SEO Works’ Instagram account between December 2024 and June 2026.

Channel profile. The account is a professional Instagram profile for a UK-based digital marketing agency, posting a mix of client results, industry commentary, and internal culture content aimed at marketing leaders, business owners, and prospective clients evaluating the agency’s expertise.

Metrics and definitions. Views are the number of times a video was played. Interactions are the likes, comments, shares, and saves recorded against the post. Engagement rate is interactions divided by views, calculated per video and then averaged across each length or topic bracket.

Controlled variables. Videos were grouped by exact runtime in seconds into half-minute length brackets, and separately by primary content topic: culture and events, client results, predictions and trend commentary, team profiles, practical tips, and a mixed “other” category covering offers and general updates.

Analytical process. Data was exported directly from Instagram’s native insights, then aggregated to calculate mean views, interactions, and engagement rate per bracket. No posts were excluded from the sample.

Data source

Download the underlying data here.